多线程FFmpeg转码Icecast2分流：如何将各进程输出重定向至独立文件？

针对每个FFmpeg进程输出独立日志的需求，有以下几种可行方案：

方案一：通过ffmpeg-python的run()方法直接重定向输出 #

修改代码逻辑，为每个电台创建独立日志文件，在调用stream.run()时将stdout和stderr定向到该文件。这种方式最简单直接，无需额外依赖。

修改后的代码示例：

import ffmpeg, csv
from threading import Thread

# 假设这些全局变量已提前定义
ICECAST2_USER = "your_user"
ICECAST2_PASS = "your_pass"
ICECAST2_PORT = "8000"
BITRATE = "64k"
SAMPLE_RATE = 44100
FORMAT = "mp3"
CODEC = "libmp3lame"

def start_stream(name, mount, source):
    icecast = f"icecast://{ICECAST2_USER}:{ICECAST2_PASS}@localhost:{ICECAST2_PORT}/{mount}"
    stream = (
        ffmpeg
        .input(source)
        .output(
            icecast,
            audio_bitrate=BITRATE, sample_rate=SAMPLE_RATE, format=FORMAT, acodec=CODEC,
            reconnect="1", reconnect_streamed="1", reconnect_at_eof="1", reconnect_delay_max="120",
            ice_name=name, ice_genre=source
        )
    )
    # 生成安全的日志文件名（替换特殊字符避免创建失败）
    log_filename = f"{name.replace('/', '_').replace(' ', '_')}_ffmpeg.log"
    # 以追加模式打开日志文件，重定向FFmpeg输出
    with open(log_filename, 'a', encoding='utf-8') as log_file:
        stream.run(stdout=log_file, stderr=log_file, overwrite_output=False)

with open('stations.csv', mode='r') as data:
    for station in csv.DictReader(data):
        thread = Thread(
            target=start_stream,
            args=(station['name'], station['mount'], station['url'])
        )
        thread.start()

方案二：用subprocess捕获输出并写入日志 #

如果需要更灵活的日志处理（比如实时监控、按日期切割日志），可以将ffmpeg命令导出为列表，用subprocess.Popen启动进程，实时读取输出并写入文件。

示例代码：

import ffmpeg, csv, subprocess
from threading import Thread

# 全局变量同方案一

def start_stream(name, mount, source):
    icecast = f"icecast://{ICECAST2_USER}:{ICECAST2_PASS}@localhost:{ICECAST2_PORT}/{mount}"
    stream = (
        ffmpeg
        .input(source)
        .output(
            icecast,
            audio_bitrate=BITRATE, sample_rate=SAMPLE_RATE, format=FORMAT, acodec=CODEC,
            reconnect="1", reconnect_streamed="1", reconnect_at_eof="1", reconnect_delay_max="120",
            ice_name=name, ice_genre=source
        )
    )
    # 导出FFmpeg命令列表
    cmd = stream.compile()
    log_filename = f"{name.replace('/', '_').replace(' ', '_')}_ffmpeg.log"
    
    with open(log_filename, 'a', encoding='utf-8') as log_file:
        # 启动子进程，合并stdout和stderr
        proc = subprocess.Popen(
            cmd,
            stdout=subprocess.PIPE,
            stderr=subprocess.STDOUT,
            text=True,
            bufsize=1  # 行缓冲，保证实时输出
        )
        # 逐行读取并写入日志
        for line in proc.stdout:
            log_file.write(line)
            log_file.flush()  # 立即写入磁盘，避免缓冲区积压
        proc.wait()

with open('stations.csv', mode='r') as data:
    for station in csv.DictReader(data):
        thread = Thread(
            target=start_stream,
            args=(station['name'], station['mount'], station['url'])
        )
        thread.start()

方案三：利用FFmpeg原生日志参数 #

如果你的FFmpeg版本支持-logfile参数（通常4.0+版本可用），可以直接在输出配置中指定日志文件路径，让FFmpeg自行写入日志。

示例代码：

import ffmpeg, csv
from threading import Thread

# 全局变量同方案一

def run(name, mount, source):
    icecast = f"icecast://{ICECAST2_USER}:{ICECAST2_PASS}@localhost:{ICECAST2_PORT}/{mount}"
    log_filename = f"{name.replace('/', '_').replace(' ', '_')}_ffmpeg.log"
    stream = (
        ffmpeg
        .input(source)
        .output(
            icecast,
            audio_bitrate=BITRATE, sample_rate=SAMPLE_RATE, format=FORMAT, acodec=CODEC,
            reconnect="1", reconnect_streamed="1", reconnect_at_eof="1", reconnect_delay_max="120",
            ice_name=name, ice_genre=source,
            loglevel="info",  # 指定日志级别，可选debug/info/warn等
            logfile=log_filename  # 指定日志文件路径
        )
    )
    return stream

with open('stations.csv', mode='r') as data:
    for station in csv.DictReader(data):
        stream = run(station['name'], station['mount'], station['url'])
        thread = Thread(target=stream.run)
        thread.start()

注意事项 #

• 日志文件名要避免特殊字符（比如/、空格），防止创建文件失败
• 如果需要长期运行，建议定期清理或归档日志文件，避免磁盘空间占用过高
• 方案二的实时输出方式需要处理子进程退出逻辑，避免线程僵死

内容的提问来源于stack exchange，提问作者Marcin Kamienski